基于实时视觉分析算法的智能图像传感器系统设计  

Intelligent Image Sensor System Design Based on Real-Time Visual Analysis Algorithms

作　　者：李志华[1] 陈耀武[1] 

机构地区：[1]浙江大学数字技术及仪器研究所,杭州310027

出　　处：《传感技术学报》2008年第7期1178-1183,共6页Chinese Journal of Sensors and Actuators

基　　金：国家863高技术研究发展计划软件重大专项资助(2003AA1Z2130);浙江省科技计划重大科技攻关项目资助(2005C11001-02)

摘　　要：设计了一种智能交通图像传感器系统以实现对监控场景的快速移动侦测和对象识别。该系统具有有线以太网和无线GPRS双重网络接入功能,硬件由基于Au1200嵌入式处理器的网络接口端和基于BlackFin 533 DSP处理器的图像分析端组成。软件系统包括运行于Au1200处理器上的基于嵌入式Linux架构的网络收发软件和运行于BlackFin 533 DSP上的视觉分析算法。本系统引入了基于区域分割的背景模型和基于特征的对象识别算法。实验结果表明该系统能够实时高效地进行自动移动检测和对象分类识别。This paper introduces a smart traffic sensor system to achieve rapid motion detection and object recognition. The system has the dual network accessing capability to the wired Ethernet and the wireless GPRS. The system hardware consists of a network interface section based on the embedded processor Aul200 and an image analysis section based on BlackFin 533 DSP. The system software includes the net- work transceiver software based on embedded Linux running in Aul200 and the visual analysis algorithms running in BlackFin 533. The sensor system utilizes a background model based on region segmentation and a feature-based object classification algorithm. Experimental results show that the system can achieve real- time automatic motion detection and object classification efficiently.

关 键 词：智能交通监控 移动侦测 对象分类 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TN911.73[自动化与计算机技术—控制科学与工程]